Research map / Label-efficient and robust learning / Recognition and long-tail
Domain Generalization via Shuffled Style Assembly for Face Anti-Spoofing
Zhuo Wang, Zezheng Wang, Zitong Yu, Weihong Deng, Jiahong Li, Tingting Gao, Zhongyuan Wang
Check your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use SSAN-M instead of the strongest listed baseline | FAS cross-dataset evaluation · M&I to C; M&I to O | HTER(%): 31.06 -> 30.00; 30.73 -> 29.44 | Improved |
| Use SSAN-R instead of SSDG-R | FAS cross-dataset evaluation · O&C&I to M; O&M&I to C; O&C&M to I; I&C&M to O | HTER(%): 7.38 -> 6.67; 10.44 -> 10.00; 11.71 -> 8.88; 15.61 -> 13.72 | Improved |
| Use SSAN-R instead of ResNet18 | large-scale FAS benchmark · Protocol 1; Protocol 2_1; Protocol 2_2 | TPR@FPR 0.1%(%): 69.10$\pm$34.34 -> 78.45$\pm$31.98; 3.64$\pm$3.93 -> 6.58$\pm$5.56; 19.00$\pm$14.79 -> 31.61$\pm$23.33 | Improved |
| Add adversarial domain alignment to content features (ablation) | SSAN-M · O&M&I to C | HTER(%): 24.44 -> 16.67 | Improved |
| Add contrastive learning for shuffled style assemblies (ablation) | SSAN-M · O&M&I to C | HTER(%): 17.59 -> 16.67 | Improved |
| Add stop-gradient to self-assembly anchors (ablation) | SSAN-M · O&M&I to C | HTER(%): 20.93 -> 16.67 | Improved |
| Use relative soft supervision instead of hard supervision (ablation) | SSAN-M · O&M&I to C | HTER(%): 28.89 -> 16.67 | Improved |
| Use shuffled auxiliary assemblies for contrastive learning instead of SCL on self-assemblies (ablation) | SSAN-M · O&M&I to C | HTER(%): 23.70 -> 16.67 | Improved |
Closest papers by meaning
- Test-Time Domain Generalization for Face Anti-Spoofing (CVPR 2024 · similarity 0.93)
- Instance-Aware Domain Generalization for Face Anti-Spoofing (CVPR 2023 · similarity 0.93)
- Rethinking Domain Generalization for Face Anti-Spoofing: Separability and Alignment (CVPR 2023 · similarity 0.92)
- Gradient Alignment for Cross-Domain Face Anti-Spoofing (CVPR 2024 · similarity 0.91)
- CFPL-FAS: Class Free Prompt Learning for Generalizable Face Anti-spoofing (CVPR 2024 · similarity 0.91)
- Rethinking Generalizable Face Anti-spoofing via Hierarchical Prototype-guided Distribution Refinement in Hyperbolic Space (CVPR 2024 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
